Focusrite Scarlett Solo Problems

[EEVblog]

Dave found two issues with his Focusrite Scarlett Solo.
Conducted mode USB noise when you unbalance the monitor outputs.
And the headphone jack not muting the monitor outputs.
Plus a demonstration on the importance of using balanced audio leads.

[Sgt_Baker]

Hi Dave,

What you're inadvertently picking up isn't "coupled" in the conventional EMI sense.  It's actual power filth on the 5V USB supply from the motherboard and is coming to prominence on account of some grounding issue/mismatch/loop/Murphy on the analog side of things.

Gimme a min - am just recording some examples of how different USB power sources can wreak havoc on USB-powered audio kit.  Back in a few.

--Baker

[EEVblog]

What you're inadvertently picking up isn't "coupled" in the conventional EMI sense.  It's actual power filth on the 5V USB supply from the motherboard and is coming to prominence on account of some grounding issue/mismatch/loop/Murphy on the analog side of things.

Yes, that's called conducted mode EMI as I mentioned.

[Sgt_Baker]

Hello again,

Disclaimer:  Not an electrical engineer.  Everything based on 20+ years audio-visual engineering experience in broadcast/film/recording studio.  Not trying to teach anybody how to suck eggs.  All provided such that somebody facing similar problems might have an “ah ha!” moment, potentially saving further headache.

TL;DR:  Source of USB power can introduce catastrophic interference to USB-powered analog audio devices dependent on ground?/earth?/0V?/5V? referencing between connected devices and USB power sources.

The following are real-world examples of several forms of audio interference which may occur when utilising USB power at one or more points in an audio signal chain.  The only factor changing between these recordings is the source of 5V USB power.  Everything else remains entirely consistent between takes.

This setup is not intended to precisely replicate Dave Jones’s problem, instead a quick-and-dirty demo that I happened to have lying around.


Playback:

Really, desperately cheap USB-powered analog PC speakers.  Can detect power supply interference on Mars and beyond.  Power via a USB-A and audio via a 3.5mm jack.  No volume control, so permanently pegged at 100% volume.  Why?  Because they exhibit precisely the problem of interest out of the box.

Audio from the rear 3.5mm speaker jack of a Gigabyte Aorus x570 Master motherboard.


Recording:

Behringer ECM8000 reference mic positioned almost in contact with speaker cone.

Focusrite Scarlett 2i2 (coincidentally) at 192KHz/24bit.  In a known-good and well-used configuration that is completely free from the type of interference we’re interested in.


Test 1 -  Baseline

Baseline.mp3

No power to speakers.  Ambient noise only.  Quiet ticking of clock.  Breathing of nearby human.  Perhaps a distant seagull.


Test 2 – Motherboard USB

Motherboard 5V.mp3

It doesn’t matter which USB socket, rear or front nor USB 2/3/3.1. They’re all the same in this regard despite drawing power from different parts of the motherboard internally.

Result:  The unbearable sound of the computer “thinking”.  Precisely that Dave records.


Test 3 – Anker 6-way mains USB power brick.

Anker 5V.mp3

Actions:  During the test we repeatedly plug/unplug a phone which draws an additional 1.73A.

Result:  Loud SMP switching noise when power bank is under low load.  Noise disappears (ultrasonic?) when the phone’s 1.73A demand is added.


Test 4 – “Good” mains USB phone charger.

Good USB 5V.mp3

Came with a long-forgotten HTC phone according to the markings.  Much less substantial than modern phone chargers and of unknown age.

Result:  Much less discernible interference.  Reason unknown.  Perhaps permanently-high/ultrasonic switching frequency?


Test 5 – USB battery bank

Battery Bank 5V.mp3

Small mophie battery bank of unknown age. Looks and feels like a 2x 18650 effort.

Result: Again, something there, but as per the "good" charger test in that it's no longer aggressively disturbing.


Test 6 – Optionally-powered USB3 hub #1

Powered USB Battery PC Switchover.mp3

Actions: During the test we repeatedly apply and remove external power supplied by the mophie power bank.

Special note:  Keyboard and mouse are also plugged in to this hub.

This one is interesting.  The hub is optionally-powered in that it will draw current from the Motherboard unless 5V is provided to a micro-USB power input.

Result:  Clear difference in that motherboard noise in greater evidence util battery 5V is applied, at which point interference is quieter, yet remains.  Potential digital crosstalk/leakage within the hub itself?


Test 7 - Optionally-powered USB3 hub #2 on 5V battery power

Powered USB Battery changing keyboard lights.mp3

The connected keyboard has a fancy per-key dimmable backlight.  Can this affect power supplied to speakers via the hub?

Actions: Switching backlight between four factory default off/low/med/high presets.

Result: Yes.

[Sgt_Baker]

Apologies.  Not trying to split hairs where nomenclature and/or underlying science is concerned.

What you're inadvertently picking up isn't "coupled" in the conventional EMI sense.  It's actual power filth on the 5V USB supply from the motherboard and is coming to prominence on account of some grounding issue/mismatch/loop/Murphy on the analog side of things.

Yes, that's called conducted mode EMI as I mentioned.

[Sgt_Baker]

Questions/quandaries for the experts:

1)  Why does the self-powered/self-signalled (both 3.5mm jack and USB power plugged in to adjacent sockets on the motherboard) example configuration counterintuitively result in the most "CPU-thinking" interference?  Common sense/experience would suggest the diametric opposite, no, owing to common reference?

2)  The Focusrite Scarlett 2i2 (thus presumably the Scarlett Solo) is a USB 2.0 device utilising a mere two pairs of a USB-C connector with a multitude of additional conductors and signal pairs "unused".  Is there the potential for misinterpretation of standards at motherboard or device end and/or those unused conductors flapping around in the breeze?  No, because the other end is a USB-A, durr Baker.

3)  Could the demonstrated interference issues have gone unheard yet been great enough to interfere with Dave's Røde box to an extent whereby they would occasionally overwhelm USB's error correction, causing the dropouts?  (This sort of thing, in my experience, is deeply dependent on how well the drivers handle wire-level comms problems.  Needless to say purchasing a very expensive Avid (your mileage may vary) thingy practically guarantees a lifetime of driver/hardware/host-related problems.

4)  The USB specifications are hyper-specific.  Are manufacturers/implementers, from cheap to expensive, merely "getting away with the bare minimum" for their product to function as expected without anybody asking too many questions?

--Baker

[David Hess]

I would like to see Dave do the same tests with a USB isolator installed.  In reference to question 3 below, a USB isolator might also solve the RODE crashing issue.

1)  Why does the self-powered/self-signalled (both 3.5mm jack and USB power plugged in to adjacent sockets on the motherboard) example configuration result in the most "CPU-thinking" interference?  Common sense/experience would suggest the diametric opposite, no, owing to common reference?

Being physically adjacent does not mean that they are electrically adjacent, and good motherboards use special layouts on their analog audio ports to reduce noise which could not be applied to a USB port.

3)  Could the demonstrated interference issues have gone unheard yet been great enough to interfere with Dave's Røde box to an extent whereby they would occasionally overwhelm USB's error correction, causing the dropouts?  (This sort of thing, in my experience, is deeply dependent on how well the drivers handle wire-level comms problems.  Needless to say purchasing a very expensive Avid (your mileage may vary) thingy practically guarantees a lifetime of driver/hardware/host-related problems.)

That could be the case.  I have seen it happen with other interfaces but never specifically with USB, but being aware of the problem, I avoid using USB where this could be an issue.

4)  The USB specifications are hyper-specific.  Are manufacturers/implementers, from cheap to expensive, merely "getting away with the bare minimum" for their product to function as expected without anybody asking too many questions?

USB is just a horrible specification for chassis to chassis audio (and test instrument) connections because of the potential for creating ground loops.  It was never designed or intended for this.

[Sgt_Baker]

USB is just a horrible specification for chassis to chassis audio (and test instrument) connections because of the potential for creating ground loops.  It was never designed or intended for this.

Moreover the USB specification is an awful choice for anything approaching "real time" as it inherently depends on the host-whatever polling downstream devices.  There exists no mechanism for a downstream device to signal the host.  Current specifications are limited to a maximum rate of 1000Hz host-to-everything-downstream-on-this-host-port polling under ideal circumstances.  And said polling is performed either at driver level or user level, depending on the device.  There is no concept of an interrupt or DMA or any other age-old concept when utilising USB.

Whichever way you're left in a situation where real-time devices are forced to deal with a jittery polling system at best, or a polling system which completely ignores the device if the host is too busy performing other tasks.  It's a minor miracle that the Focusrite and Røde work at all given their low latencies.

[Sgt_Baker]

Incidentally, the Focusrite Scarlett devices present a plurality of devices to the host.  One being the device itself, and another being a USB Mass Storage Device containing the latest iteration of the kernel-level drivers at the time of the device's manufacture. Thus, when you first plug the Scarlett in, it self-presents its own drivers to Windows on a "USB thumb drive" before enabling the audio-streaming aspects of the system on account of said driver attempting to poll the device.  "If device not polled, present thumb drive. Etc."  So Dave's assertion that "it's a standard windows audio device" is technically incorrect.

[HattedSquirrel]

So Dave's assertion that "it's a standard windows audio device" is technically incorrect.

Thats not fully correct either. The Scarletts have several modes of operation. In one they actually present themselves as Standard USB Audio Device for which every modern operating system has drivers onboard. This mode is the only reason they work under Linux and even on Android. It is possible to end up using this mode in Windoes, too. Happened to me.
The intended use, however, is in conjunction with their own set of drivers and control software. If that is present and installed on the PC, it'll switch into a mode with more capabilities.

When I first connected my Scarlett to my Windows machine, I saw the USB Mass Storage Mode you described, but it only promped me to install a new firmware and opened a browser window. The Focusrite control software did not get installed. I then quickly found out that I had to install that stuff manually if I actually wanted to do something useful.

The 18i20 has an "alt monitor" button, but I believe they cheaped out on that for the solo. It /might/ exist in the control software, though.

[Sgt_Baker]

In that case I stand corrected, as always.

[EEVblog]

So Dave's assertion that "it's a standard windows audio device" is technically incorrect.

Thats not fully correct either. The Scarletts have several modes of operation. In one they actually present themselves as Standard USB Audio Device for which every modern operating system has drivers onboard. This mode is the only reason they work under Linux and even on Android. It is possible to end up using this mode in Windoes, too. Happened to me.
The intended use, however, is in conjunction with their own set of drivers and control software. If that is present and installed on the PC, it'll switch into a mode with more capabilities.

Correct. I installed the driver as people said it had option to do what I wanted. Nope, it doesn't. It just lets you update the firmware (whcih didn't seem to work BTW) and chaneg the LED colours, and default operation of the 48V button. Output mixing options are not supported on the Solo. So I removed it and the driver, just using the default windows driver again.

[EEVblog]

I would like to see Dave do the same tests with a USB isolator installed.  In reference to question 3 below, a USB isolator might also solve the RODE crashing issue.

I doubt it, because:
1) It actually work just fine as a monitor output, it's just the microphone that drops out.
2) It has detection issues with Zencaster, yet no other problems with other programs and browser sites.
3) This problem has persisted across different processors, USB ports and cables etc.

[Sgt_Baker]

Hi all,

In the interests of completeness it's worth noting that device driver and Focusrite Control are two (almost) completely independent entities performing completely different tasks.  One allows Windows/OS X/Linux to communicate with the hardware at magic-over-usb level, whereas the other is hilariously pointless in the two-channel models unless, as observed, one exhibits an obsession with purple peaking LED rings or you wish to switch between line/instrument impedance or "Air" (an EQ to make vocals sound "better") in software as opposed to pressing the button on the unit.

But on to practical matters.  Are we all missing a trick here?  In the video Dave states that the Røde would mute the speaker outputs when inserting a jack in to the headphone socket.  Surely the same effect is achieved on the Focusrite by merely turning the Monitor pot to zero?

[EEVblog]

In the interests of completeness it's worth noting that device driver and Focusrite Control are two (almost) completely independent entities performing completely different tasks.  One allows Windows/OS X/Linux to communicate with the hardware at magic-over-usb level, whereas the other is hilariously pointless in the two-channel models unless, as observed, one exhibits an obsession with purple peaking LED rings or you wish to switch between line/instrument impedance or "Air" (an EQ to make vocals sound "better") in software as opposed to pressing the button on the unit.

Which is exactly what I found out after everyone screamed at me to installed the software 

But on to practical matters.  Are we all missing a trick here?  In the video Dave states that the Røde would mute the speaker outputs when inserting a jack in to the headphone socket.  Surely the same effect is achieved on the Focusrite by merely turning the Monitor pot to zero?

Because the monitor volume control works for both monitor and headphone volume at once.
This is thing people fail to understand with this. What is the point of having a device with a single volume control that works on both speakers and headphones and doesn't disable the monitors when the headphones are plugged in. It's kinda stupid, hence why Rode doesn't do that.

[themadhippy]

Wonder if the rhode box uses  proper balancing transformers whilst the focusrite is "electronically" balanced 

[EEVblog]

Wonder if the rhode box uses  proper balancing transformers whilst the focusrite is "electronically" balanced

No, you can see this in the teardown video.

[SiliconWizard]

And the headphone jack not muting the monitor outputs.

About that, just a thought but: does the doc claim that should happen? If so obviously, this is a defect. If not... well, I've had quite a few external audio interfaces over the years (most of them "semi-pro" gear) and AFAIR, none of them ever had this feature. Actually, in studio environments, that plugging headphones would mute monitor outputs could be extremely annoying.

[EEVblog]

3)  Could the demonstrated interference issues have gone unheard yet been great enough to interfere with Dave's Røde box to an extent whereby they would occasionally overwhelm USB's error correction, causing the dropouts? 

No, because the problem extend to not being recognised by web based applications, but being detected fine by other local programs. So that's a compatibility thing, and one that changed over time as Zencaster was developed for example.
And the USB never "dropped out", as the monitor side always work fine all the time, it's just that the microphone would output nothing.

[EEVblog]

And the headphone jack not muting the monitor outputs.

About that, just a thought but: does the doc claim that should happen? If so obviously, this is a defect. If not... well, I've had quite a few external audio interfaces over the years (most of them "semi-pro" gear) and AFAIR, none of them ever had this feature. Actually, in studio environments, that plugging headphones would mute monitor outputs could be extremely annoying.

Sure, but when you have a SINGLE volume control that control BOTH headphones and monitor volume as the same time, and CANNOT be independetly varied, what use is having both on at the same time?
Maybe if you have seperate volume on your monitors speakers, ok, or you have some other way to vary the monitor volume, but then you aren't going to use an inflexible box like this to begin with.
I content that is makes no sense at all to not disable the output, which is why Rode does it.
I knew the 2i2 model had indepent controls, whcih is why I assume the single control Solo would work like the Rode does and disable to output. But it doesn't, and I think that is dumb on a unit where you can't vary the otuput of both being driven.

[SiliconWizard]

And the headphone jack not muting the monitor outputs.

About that, just a thought but: does the doc claim that should happen? If so obviously, this is a defect. If not... well, I've had quite a few external audio interfaces over the years (most of them "semi-pro" gear) and AFAIR, none of them ever had this feature. Actually, in studio environments, that plugging headphones would mute monitor outputs could be extremely annoying.

Sure, but when you have a SINGLE volume control that control BOTH headphones and monitor volume as the same time, and CANNOT be independetly varied, what use is having both on at the same time?

Yes, if the interface doesn't allow you to control them independently, then I agree it doesn't make sense.

[SiliconWizard]

So Dave's assertion that "it's a standard windows audio device" is technically incorrect.

Thats not fully correct either. The Scarletts have several modes of operation. In one they actually present themselves as Standard USB Audio Device for which every modern operating system has drivers onboard. This mode is the only reason they work under Linux and even on Android. It is possible to end up using this mode in Windoes, too. Happened to me.
The intended use, however, is in conjunction with their own set of drivers and control software. If that is present and installed on the PC, it'll switch into a mode with more capabilities.

I do not have this interface so I can't take a look, but for the audio streaming part alone, I would be surprised if Focusrite didn't just use USB Audio Device anyway, even with their own drivers?

Now we may have to consider which OSs it's compatible with. Windows didn't get native USB Audio *2.0* (which would natively support the higher sampling rates, bit depth, etc) until Windows 10 (yes!! if I'm not mistaken...) - and maybe not even the earlier builds - so if it has to support older versions of Windows, it would require non-native drivers, but they may still implement USB Audio 2.0, which is more than likely, since the interface uses XMOS chips and as I said in the other thread, XMOS provides a ready-to-use USB Audio 2.0 implementation...

So that would just explain the need for their own drivers. On older Windows versions, only USB Audio 1.0 is supported, and that's probably what the interface falls back to when they are used with such Windows versions and the Focusrite drivers haven't been installed yet. No need to say that in this mode, you won't get the full performance.

USB Audio 2.0, OTOH, has been supported by Linux for a pretty long time now. So, just to make things clear.
If anyone has this interface and a Linux install, they can just have a look. Plug it in, and look at dmesg output. It's very likely an USB Audio *2.0* device, plus probably additionally one or more HID devices for controlling the interface.

[Sgt_Baker]

3)  Could the demonstrated interference issues have gone unheard yet been great enough to interfere with Dave's Røde box to an extent whereby they would occasionally overwhelm USB's error correction, causing the dropouts? 

No, because the problem extend to not being recognised by web based applications, but being detected fine by other local programs. So that's a compatibility thing, and one that changed over time as Zencaster was developed for example.
And the USB never "dropped out", as the monitor side always work fine all the time, it's just that the microphone would output nothing.

A what now?

[EEVblog]

3)  Could the demonstrated interference issues have gone unheard yet been great enough to interfere with Dave's Røde box to an extent whereby they would occasionally overwhelm USB's error correction, causing the dropouts? 

No, because the problem extend to not being recognised by web based applications, but being detected fine by other local programs. So that's a compatibility thing, and one that changed over time as Zencaster was developed for example.
And the USB never "dropped out", as the monitor side always work fine all the time, it's just that the microphone would output nothing.

A what now?

The Focusrite works fine. I just need to turn off the speakers when I plug in the headphones.